Abstract
The present study compared the effectiveness of a self-affirmation and a role model guest lecture intervention on reducing students’ perceptions of science-related social identity threat. Participants included 67 Latino high school students enrolled in a college preparation program. Students were randomly assigned either to a self-affirmation intervention or a self-affirmation control task, and the role model intervention was open to all students, with some choosing to participate. Results from an ANCOVA found the combination of both interventions had an identity threat reducing effect of moderate magnitude on perceptions of identity threat, and planned contrasts found statistically significant differences in perceptions of identity threat between students who received both interventions and no intervention, and between students who received both interventions and the self-affirmation task alone. Our research suggests that using multiple and combined interventions might provide an important advantage in order to reduce perceptions of identity threat in Latino students.
Latinos are the largest and fastest growing ethnic minority group in the United States (Colby & Ortman, 2015), but despite this demographic trend, they remain underrepresented in STEM (science, technology, engineering, and mathematics) fields (National Science Foundation, 2016). Researchers have proposed a variety of hypotheses to explain the underrepresentation of Latino students in STEM fields, including the impact of psychosocial processes on students’ educational experiences.
Latino students in the United States have to cope with an array of negative stereotypes about their academic success (Gonzalez, Blanton, & Williams, 2002), their intelligence (Gandara & Contreras, 2009), and their ability to succeed in STEM fields (Gandara & Contreras, 2009). The prevalence of these stereotypes puts students at risk of experiencing the detrimental effects of social identity threat. Social identity threat theory posits that members of underrepresented groups experience a variety of negative psychosocial effects related to their status as minorities, whether it is gender, race, sexual orientation, or ethnicity. Whenever a minority status places an individual in a position in which he or she feels devalued due to their identity, this individual might experience social identity threat (Crocker, Major, & Steele, 1998; Major & O’Brien, 2005). Instances of overt discrimination have been shown to strongly impact the academic performance of some minority students, a phenomenon known as stereotype threat (Steele, 1997). Stereotype threat is believed to impact students’ performance in achievement settings through a dual process (Schmader, Johns, & Forbes, 2008). When students are confronted with discrimination in an educational context, they become immediately stressed. Simultaneously, their monitoring processes activate, and they begin to engage their cognitive resources in a variety of tasks aimed at making sense of the threatening situation and coping with the physiological stress responses they are experiencing. As a result, the cognitive and emotional control processes that should be used in the academic task are diverted by the discrimination event, which then disturbs students’ performance by intervening with their working memory processes. This hypothesis has been supported in studies conducted with Latino students in which students were threatened by having the researchers frame their achievement task as a measure of their intelligence (Gonzalez et al., 2002; Schmader & Johns, 2003).
Social identity threat is assumed to operate through very similar processes. According to Steele, Spencer, and Aronson (2002), environmental cues that signal threat prompt individuals to engage in a process of cognitive vigilance. In this stage, individuals under threat are looking for cues that can confirm the likelihood of being devalued in that environment, while also acting under a strong motivation to ignore the evidence of possible discrimination. This tension creates what Steele et al. (2002) refer to as “ruminative conflict” (p. 419), a cognitively taxing exercise that distracts their attention and might eventually pressure individuals to disengage from the environment or domain in question. Unlike stereotype threat, individuals do not need to experience a blatant discrimination event that consciously activates the stereotype in order to experience identity threat (Crocker et al., 1998, Inzlicht & Good, 2006; Murphy, 2007; Murphy, Steele, & Gross, 2007). Experiences of identity threat might be elicited in environments where members of minority groups find themselves to be numerically underrepresented in a particular setting, in workplaces or academic institutions that follow very strict and unique set of values, or in which chances for improvement or advancement are ruled by unclear guidelines (Steele et al., 2002). As stated by Sherman et al. (2013), “merely being aware that one’s social or group identity could cause one to be devalued can prove psychologically threatening” (p. 592).
Self-affirmation interventions have been shown to reduce the negative effects of identity threat by having students affirm core values about themselves (Cohen, Garcia, Apfel, & Master, 2006; Cohen, Garcia, Purdie-Vaughns, Apfel, & Brzustoski, 2009; Cook, Purdie-Vaughns, Garcia, & Cohen, 2012; Sherman et al., 2013; Yeager & Walton, 2011). According to Sherman et al. (2013), when students write about their moral, personal, or religious values, they strengthen their self-concept, which then reduces the stressful effects of identity threat and allows them to better cope with situational destabilizing events in the environment. As stated by the authors, “affirmations can help shore up the ‘narrative of adequacy’ that students form as they make sense of their academic struggles, and this can prevent a negative event from threatening the self” (Sherman et al., 2013, p. 594). Cook et al. (2012) also propose that self-affirmations allow individuals to encode threatening situations as a general threat, rather than as an indication of their personal self-worth. Henceforth, self-affirmations allow students not only to anchor their worth in other aspects of the self (Sherman et al., 2013) but also to contextualize their experiences of identity threat and separate their perceptions of belonging from it (Cook et al., 2012). The benefits of these self-affirming tasks are evidenced in the long-term increase of minority students’ academic performance (Cohen et al., 2006; Cohen et al., 2009; Sherman et al., 2013) and have been replicated with Latino middle school students (Sherman et al., 2013).
Self-affirmation interventions show how concentrating on other aspects of the self can help students reframe and cope with threat. Dasgupta (2011) points out that while many of the interventions aimed at reducing the effects of threat “attempt to change perceivers’ inner appraisals or construals of the self, the in-group, or ability domain” (p. 235), directly changing situational factors and exposing students to positive in-group models of success also constitute a way to disrupt the effects of threat on students who are numerically underrepresented in academic settings such as STEM fields.
The stereotype inoculation model posits that exposure to peers and role models can significantly impact the academic and career trajectory, and educational attainment of students by strengthening their self-concept, increasing their sense of belonging, their self-efficacy, and improving their attitudes toward domains in which they are underrepresented (Dasgupta, 2011). The impact of these role models is hypothesized to be particularly beneficial for students who are at risk of experiencing threat due to their lack of numerical representation in a particular academic domain (Murphy et al., 2007), whose identities are at odds with the values held in their academic or work environment (Purdie-Vaughns, Steele, Davies, Ditlmann, & Crosby, 2008) or those who highly identify with a domain in which negative stereotypes exist about their group (Schmader et al., 2008). Dasgupta (2011) argues that exposure to experts who belong to the same group can provide students with high achieving models, in a way that “inoculates individuals against self-doubt” (p. 233) and opens to these students the opportunity to succeed in academic or career fields they might have never considered before. Exposure to role models can also impact the way individuals apply pervasive societal stereotypes to themselves (Stout, Dasgupta, Hunsinger, & McManus, 2011). However, several factors, like role models’ deservingness, can play a role on their ability to dismantle the effect of stereotype threat on others. McIntyre, Paulson, Taylor, Morin, and Lord (2011) demonstrate that only those role models who were perceived as worthy and deserving of their success, due to stable and internal attributes, were able to reduce the effects of stereotype threat in a group of women taking a math test. Another factor that might influence this process is the model’s ability to show achievement as malleable, and intelligence as incremental. Manke and Cohen (2011) emphasize how important it is for role models to convey their own stories of struggle and allow students to see how overcoming obstacles constitutes an intrinsic part of learning and succeeding. They also argue that, similar to self-affirmations, exposure to these expert role models could affirm the self by fostering a positive image of the individual’s group in a threatening domain. This assertion is consistent with findings purported by research conducted with group affirmations (Derks, van Laar, & Ellemers, 2009).
While different interventions have shown to positively impact perceptions of social identity threat and its associated negative effects, there is still a lack of research that explores the added value of using combined strategies to reduce the effects of identity or stereotype threat (Yeager & Walton, 2011). Researchers have also advocated for the necessity of testing the usefulness of such interventions through longitudinal studies in the field (Dasgupta, 2011), outside the control of the lab manipulation (Shapiro, Williams, & Hambarchyan, 2013), and in educational environments where the effects of psychosocial interventions intertwine with other complex learning and social processes (Yeager & Walton, 2011).
In our present study, we compared the effectiveness of a self-affirmation intervention and a role model intervention in a group of Latino students enrolled in a college preparation program. This research explored the impact of both interventions on Latino students’ perceptions of identity threat in this stereotyped domain.
Method
Participants
The participants in the current study were 67 high school students enrolled in a college preparation program (see Table 1). This program is part of a collaboration between a Southeastern research university and a local school district, and is aimed at increasing Latino K-12 students’ STEM interest and strengthening the pipeline of Latino students into postsecondary STEM education.
Participant Demographics by Intervention Condition.
As part of the program, the participants receive college mentoring and academic tutoring once per week and after school by undergraduate student mentors from a variety of local universities. This is a cohort-based longitudinal program, and once students enroll, the goal is to retain them until graduation. Students are recruited by program staff based on teacher recommendations. In order to be enrolled into the program, students and their parents must complete an application, students must not have any major behavioral problems in school, and they must not have failed more than two academic classes in the previous academic year.
Once students are enrolled in the program, they are invited to participate in a longitudinal research study. As part of this study, they complete a presurvey upon program entry 1 and a postsurvey at the end of each academic year, while they stay in the program. Participants had to submit an Institutional Review Board (IRB)-approved caregiver consent and a student assent to be in the study.
Interventions
We implemented two interventions in the second year of the program designed to reduce the impact of ethnic identity threat in the group of participating Latino students. The interventions were a self-affirmation writing task and a role model intervention. Students were randomly assigned either to the self-affirmation intervention or to a self-affirmation control task. All students in the program were invited to participate in a lecture offered by a Latino Aerospace Engineer (role model intervention) and 42% of the students attended the lecture. Thus, the study was a 2 (role model intervention vs. control) × 2 (self-affirmation intervention vs. control) quasi-experimental design. Some students received the self-affirmation intervention only, some received the role model intervention only, some received both interventions, and others received no intervention. As only the self-affirmation intervention could be randomized, the group sizes are not equal. See Table 1 for the number of participants in each group.
The self-affirmation writing task sought to combat the negative effects of identity threat by having students affirm their core values. We used the previously validated self-affirmation intervention tasks designed by Cohen et al. (2006). Students in the self-affirmation intervention were asked to think about values that were important to them, why they were important to them, and how they affected their lives. In contrast, the control task asked students to think about values that were least important to them, how and why these values might be important to others, and how these values affect other people’s lives.
The second intervention was a lecture given by a successful Latino STEM professional in the field of Aerospace Engineering. During the presentation, this male speaker talked about his educational path: from his origins in another country to being a STEM professional in the United States. He talked about the challenges and stereotypes he had to overcome; how effort, hard work, and perseverance allowed him to arrive to where he is today; about prevalent issues of ethnic identity that exist nowadays in the United States; and how to combat common stereotypes about Latino students. By addressing these topics, the role model emphasized practices known to increase identification with a model: how ability can be malleable and therefore success is not driven by uncontrollable factors (Manke & Cohen, 2011; Walton & Cohen, 2007), how the future and success evoked by the model is positive (Oyserman, Destin, & Novin, 2015), and how struggling can be a common aspect of a successful educational path (Walton & Cohen, 2007). In addition, he highlighted how concerns about negative stereotypes about their group can be common for minorities, but surmountable. This last factor has the potential to make the message of struggle even more relevant to the students and therefore increase a sense of connection with the role model. The fact that these stereotypes were directly addressed by this successful model not only legitimatizes and acknowledges a real concern from the students, but, at the same time, it serves to refute and dispel those stereotypes while fostering a positive group image in the participants (Manke & Cohen, 2011). During this lecture, participants had an opportunity to share their own concerns about the stereotypes they themselves face every day. The lecture was open to all students in the program, and assignment to this intervention could not be randomized for ethical reasons. Therefore, students self-selected to this intervention. The limitations of not having a complete experimental design are addressed in the “Discussion” section.
Measures
This intervention is part of a larger longitudinal study evaluating the effectiveness of a college preparation program. The surveys completed by students enrolled in the program assessed the perceptions of science-related identity threat by asking them, “How likely do you think it is for people who share your ethnic background to become a scientist in this country?”—an adapted item from the “Is science me?” scale (Aschbacher, Li, & Roth, 2010). Response choices included “very unlikely,” “somewhat unlikely,” “somewhat likely,” “very likely,” and “not sure.” The “not sure” responses in this study were always coded as missing responses, and those responses were not included in the analysis. High scores on this question indicated low perceptions of identity threat, while low scores reflected high levels of identity threat.
Analysis
All analyses were conducted using SPSS software. Two ANCOVA tests were conducted to determine differences in postintervention scores on identity threat. We conducted one ANCOVA with pretest identity threat score as covariate to control for students’ preexisting levels of identity threat and to show the effects of the identity threat intervention on their current levels of identity threat. We also conducted an ANCOVA with preintervention identity threat as covariate to control for the effects of the program at large (i.e., additional mentoring and guidance received through program participation) and to show the effects of the specific identity threat intervention.
In the first analysis, the control variables included (1) the participants’ pretest identity threat score, (2) the number of years they have participated in the college preparation program, and (3) the interaction between intervention group by pretest score by program year to test the assumption of homogeneity of regression slopes. Participants who did not have a valid pretest and postintervention score were excluded from the analysis.
The second ANCOVA had the following control variables: (1) the participants’ preintervention identity threat score, (2) the number of years they have participated in the college preparation program, and (3) the interaction between intervention group by preintervention score by program year to test the assumption of homogeneity of regression slopes. For participants in their first program year, the preintervention score was the same as the pretest score. For all others, the preintervention score was the score from the previous year’s end of year survey (i.e., postscore from Year 1 of the longitudinal study).
ANOVA or chi-square tests were conducted to check the assumption of independence of covariate and treatment effect in both the ANCOVA in the first analysis controlling for pretest identity threat score, as well as the ANCOVA in the second analysis controlling for preintervention identity threat score. Levene’s tests were included in all ANCOVAs to test the assumption of equality of error variances.
Results
A one-way ANCOVA was conducted to determine a statistically significant difference between groups across interventions (no intervention, self-affirmation only, role model only, and both self-affirmation and role model interventions) on perceptions of identity threat. As stated before, we controlled for preprogram identity threat scores (see Table 2), years in the program, and the interaction of covariates and intervention.
Stereotype Threat Analyses Means and Standard Deviations.
ANCOVA was conducted controlling for preprogram scores, years in the program, and the interaction of covariates and the intervention.
ANCOVA was conducted controlling for preintervention scores and years in the program.
There was a statistically significant effect of the intervention on postintervention identity threat scores after controlling for the covariates, F(3, 46) = 3.198, p = .032. Planned contrasts revealed that having both interventions significantly increased students’ perceptions of how likely it is for someone from their ethnic background to become a scientist in this country compared with having no intervention, t(46) = –2.897, p = .006, r = .393 (both interventions: M = 2.86, SD = 1.10; no intervention: M = 2.38, SD = 1.07), therefore indicating an identity threat reducing effect. Both interventions were also statistically more likely to increase positive perceptions of Latinos in science compared with having only the self-affirmation intervention, t(46) = –2.314, p = .032, r = .323 (both interventions: M = 2.86, SD = 1.10; self-affirmation intervention: M = 2.33, SD = 0.78). The effect sizes, r = .393 and r = .323, reveal moderate effects for both contrasts (see Figure 1).
Stereotype threat postintervention means and standard deviations for ANCOVA controlling for preprogram scores, years in the program, and the interaction of covariates and the intervention.
A second one-way ANCOVA was conducted, this time replacing preprogram identity threat with preintervention identity threat as a covariate. There were no statistically significant differences between the intervention groups on postintervention identity threat after controlling for preintervention identity threat, years in the program, and the interaction of covariates and intervention, F(3, 42) = 2.064, p = .119. While this analysis did not have statistically significant results, we believe it might be due to issues of power. As the interaction between the covariates and the independent variable to test the homogeneity of regression slope was not significant, we reran the analysis with only the preintervention identity threat scores (see Table 2) and years in program as the covariates.
There was a statistically significant effect of the intervention on postintervention identity threat scores after controlling for the covariates (but not controlling for their interaction), F(3, 46) = 3.191, p = .032. Planned contrasts revealed similar results to the analysis controlling for preprogram identity threat, with both interventions statistically significantly reducing perceptions of identity threat compared with no interventions, t(46) = –2.542, p = .014, r = .351 (both interventions: M = 3.30, SD = 0.82; no intervention: M = 2.27, SD = 0.99), and both interventions statistically significantly reducing identity threat compared with self-affirmation alone, t(46) = –2.831, p = .007, r = .385 (self-affirmation intervention: M = 2.25, SD = 0.87). The effect sizes, r = .351 and r = .385, reveal moderate effects for both contrasts, respectively (see Figure 2).
Stereotype threat postintervention means and standard deviations for ANCOVA controlling for preintervention scores and years in the program.
Discussion
ANCOVAs showed that the combination of both interventions had an identity threat reducing effect of moderate magnitude despite the limitations of power in our study. Students who attended the role model lecture and also completed the self-affirmation task had lower perceptions of identity threat at the end of the academic year. We also found that the two interventions together statistically significantly reduced identity threat compared with the self-affirmation task alone. However, there were no statistically significant differences in perceptions of identity threat between students who received both interventions compared with those students who attended the guest lecture alone.
In our study, science-related perceptions of threat were captured by asking students to think what was the likelihood of someone from their ethnic group to become a scientist, rather than to think how likely it was for them to become a scientist. While the self-affirmation did not statistically reduce perceptions of threat, the role model intervention alone did not statically significantly reduce identity threat either. In turn, both interventions together were necessary to significantly reduce students’ science-related perceptions of threat.
There are two plausible explanations for this finding. On the one hand, threat reduction might have depended upon the use of multiple buffering messages against identity threat. On the other hand, it is possible that in this context, while the role model intervention provided a mechanism for students to positively frame the success of the group in the science domain, only when coupled with a self-affirmation exercise were students able to benefit from both strategies. Identifying a positive role model on a stereotypically negative domain might serve to derive positive self-worth from the group, but transferring that positive perception to inoculate the self might not be an inevitable process. In turn, the self-affirmation might have served in this study as the mechanism that might have allowed that transfer to happen. This finding is consistent with the notion that stereotype-reducing interventions are complex mechanisms that might work in concert to potentiate their impact, as speculated by Yeager and Walton (2011), rather than completely distinct strategies working in isolation in order to reduce distinct and not interchangeable types of threat (Shapiro et al., 2013).
Other factors can also play a role in how students process threatening situations. These include participants’ collectivistic self-construals, their level of identification with the threatened group, to what extent they consciously value that identification (e.g., sense of empowerment), and their valuation of the group (positive or negative). Our study is limited in that we did not assess these factors in our participants, as these measures were not part of the longitudinal study in which students were enrolled. Future research should explore the interrelationship of these mechanisms.
Another limitation is that due to sample size constraints and unequal samples within conditions, we were unable to perform any gender comparisons. Our STEM-focused program is unique in that most of the participants enrolled in our program are girls. As shown in Table 1, the percentage of girls in each condition is higher than boys. We should point out that the gender composition of the role model intervention only group was 100% female, even after the groups were randomly assigned to the self-affirmation task. This might have impacted the level of identification with the role model, whose gender was male. Also, it poses an interesting consideration for future studies, as females are hypothesized to be at risk for experiencing higher levels of threat due to their double minority status as females and Latinas in STEM (Gonzalez et al., 2002). Future research conducted with direct access to role models in real-life situations should also attempt to identify what characteristics of the role model interventions are more conducive to fostering a sense of connectedness with the study participants, including the interplay of ethnicity and gender, so that these interventions can be further replicated in other settings. It would also be important to explore the effect of multiple and combined interventions on educational outcomes known to be impacted by stereotype-reducing strategies such as sense of belongingness at school, self-efficacy, and academic performance.
For minority students who live in societies in which their identity is negatively perceived in a variety of domains, and for whom threatening situations are a common experience, processes of group identification might coalesce in more stable forms of thinking and action. Therefore, it is also imperative to investigate the connections between identity threat and how individuals perceive their relationship to their group, and how perceptions of control and their related self-directed emotions of pride, self-esteem, guilt, and shame (Weiner, 2005) might interplay with these mind-sets and achievement behaviors.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was conducted as part of a grant funded by The Goizueta Foundation.